Proceedings of the 2012 9th International Pipeline Conference IPC2012 September 24-28, 2012, Calgary, Alberta, Canada

IPC2012-90690

MULTI-TIER TENSILE STRAIN MODELS FOR STRAIN-BASED DESIGN PART 1 FUNDAMENTAL BASIS
Yong-Yi Wang, Ming Liu, and Fan Zhang Center for Reliable Energy Systems Dublin, OH, USA David Horsley BP Calgary, Alberta, Canada Steve Nanney PHMSA, US DOT Houston, TX, USA

ABSTRACT This is the first paper in a three-paper series on the tensile strain design of pipelines. The formulation of the multi-tier models [1] and evaluation of the models against experiment data [2] are presented in two companion papers. This paper starts with an introduction of general concept of strain-based design. The central part of the paper is then devoted to the tensile strain capacity, including (1) physical process of tensile strain failure, (2) limit states of tensile strain failure and associated toughness representation, and (3) fundamental basis of tensile strain models. The most significant part of the fundamental basis, the limit state of tensile failure and associated representation of the materials toughness, is given the greatest amount of attention. KEYWORDS Pipeline, strain-based design, girth weld integrity INTRODUCTION Role of Strain-Based Design (SBD) Under normal operation conditions, pipelines experience a higher level of stress in the hoop direction than in the longitudinal direction. Consequently, traditional pipeline designs focus on the control of the hoop stress, typically limiting its maximum value to a certain percentage of the specified minimum yield strength (SMYS). The detection and assessment of threats for in-service pipelines also assume the hoop stress being the primary driver for potential failures. Strain-based design focuses on potential failures driven by high longitudinal strains. Pipeline failures driven by longitudinal stresses or strains are relatively rare events in comparison to failures driven by hoop stresses. In most cases, those failures are associated with ground settlement/movement or other unusual upsetting events. Small ground settlement

may be the result of soil compaction, soil creep, soil saturation due to heavy rainfall, or differential settlement at tie-ins and crossings. Larger ground movement may come from landslides, frost heave or thaw settlement, mine subsidence, or earthquakes. Upheaval buckling induced by the transportation of hot fluids can impose large longitudinal strains on pipelines. The loading transmitted to pipelines is typically a combination of displacement- and load-controlled events. The resulting applied stresses/strains in the longitudinal direction can have a wide spectrum, ranging from small strains to very large strains. Offshore pipeline installation by reeling can impose longitudinal strains up to 2-3%. The loading is displacementcontrolled. The materials undergo multiple strain reversals. The post-construction state of girth welds, including material properties and remaining flaw dimensions, serves as the starting point of post-construction girth weld integrity management. It may be noted that the girth weld flaw acceptance criteria are applied prior to the installation by reeling. Any growth of flaws during the reeling installation needs to be accounted in the post-construction integrity evaluation. The common practice is limiting the flaw growth during reeling installation to a small amount. Limit States in Strain-Based Design Two limit states are considered in strain-based design: tensile rupture and compressive buckling. The tensile rupture is an ultimate limit state which is related to the breach of the pressure boundary. The compressive buckling is generally considered a serviceability limit state. However, it could be an ultimate limit state under a load-controlled loading scenario. Each limit state has a demand (strain demand or applied strain) and capacity (strain capacity or strain limit) component. The limit states can be represented in either deterministic or

1 Copyright 2012 by ASME This work is in part a work of the U.S. Government. ASME disclaims all interest in the U.S. Governments contributions.

probabilistic format. In the traditional deterministic format, a safe state is achieved when the strain demand multiplied by a safety factor of greater than 1.0 is less than the strain capacity. Overall Components of Strain-Based Design Key components of SBD and their relationship are schematically shown in Figure 1. The strain demand (alternatively termed applied strain) may be estimated from conditions causing the longitudinal strains, such as geotechnical and soil environment and operational conditions. Compressive and tensile strain capacity may be determined from their respective models and/or experimental test data. The strain demand and the strain capacity are then compared to determine if a postulated event or condition is safe. The characteristics of pipe and welds, such as the mechanical properties, dimensional tolerance and features, and flaw dimensions, can impact all components to varying degrees.
Geotechnical Condition Compressive Strain Demand Soil Condition Strain Demand Pipe Characteristics Girth Weld Characteristics Operation Condition Tensile Strain Demand

Extensive guidance is provided on fracture toughness testing. DNV OS RP F108 [11] provides further guidance on the engineering critical assessment (ECA) of girth welds for installation by reeling. The recommended toughness testing procedure is a multiple-specimen SENT (single-edge-notched tension) procedure with further qualification and validation by so-called sector specimens. The sector specimen is similar to a miniature curved wide plate specimen. DNV F101 and F108 collectively provide insightful comments related to many complex factors affecting TSC. The implementation procedures of those comments are, in many instances, not prescribed. Therefore, the application of F101 and F108 requires the involvement of well-seasoned experts. The platform of ECA, i.e., Level 3 of BS 7910 [12], is not the most suitable format for strain-based design, although it serves as a useful intermediate step before more suitable formats are developed and validated. The full implementation of F101 and F108 can be prohibitively expensive for large onshore projects. Initiation/Apparent Toughness Approach In the initiation/apparent toughness approach, the TSC is established by equating the crack driving force, i.e., CTODF, with the initiation toughness (alternatively termed apparent toughness) of the structure. The CTODF is related to the remote longitudinal strain. The relationship depends on flaw size, flaw location, linepipe tensile properties, weld metal tensile properties (weld strength mismatch), weld geometry, and other parameters [13,14,15,16,17]. The initiation/apparent toughness approach has been formally recognized in Annex C of CSA Z662 2007 Edition [18]. The significance of the apparent toughness in the physical process of tensile strain failures is further elaborated in later sections of this paper. Resistance Curve Approach (Tangency Approach) The resistance-curve approach assumes ductile instability being the point of failure. The crack driving force relation is essentially the same as that of initiation/apparent toughness approach. The resistance curve (R-curve) is directly measured from test specimens. The failure point or the unstable ductile tearing point is determined by the traditional tangency criteria. Examples of the resistance curve approach are those published by SINTEF [19,20,21] and ExxonMobil [22,23,24,25]. Osaka University and JFE Approach The Osaka/JFE approach relies on the failure loci relating the stress triaxiality and equivalent plastic strain at a crack tip. The so-called two-parameter approach has been applied to a wide variety of fracture mechanics applications [26,27]. Igi and Suzuki applied this methodology for the prediction of tensile strain limit of X80 pipes [28]. SENT and CWP specimens were tested to establish the failure loci. Igi and Suzuki demonstrated the effects of internal pressure and Y/T on tensile strain limits using this method. The reduction of tensile strain limits from non-pressurized to pressurized

Compressive Strain Capacity

Tensile Strain Capacity

Does design/ operation pass compressive strain criteria? Yes

No

No

Does design/ operation pass tensile strain criteria? Yes

Final Design

Figure 1 Key components of SBD and their relationship State of Art in Tensile Strain Capacity (TSC) Curved Wide Plate Testing Curved wide plate (CWP) testing has been one of the most recognized tools for determining girth weld tensile strain capacity [3,4]. A few notable shortcomings of past CWP tests are the lack of consistent and standardized test procedures [5] and their inability to evaluate the effects of internal pressure and girth weld high-low misalignment on TSC. Denys, et al., have published a recommended testing procedure of CWP specimens [6]. Many fundamental elements towards a consistent test procedure have been developed by a joint research team from CRES, NIST, CANMET, and Lincoln Electric [7,8,9]. Extension of Stress-Based Design Procedures Section G100 of DNV OS-F101 [10] provides guidance on the determination of girth weld defect acceptance criteria for strain-based design conditions. A number of key input parameters are covered in the guidelines, including: (1) the selection of appropriate stress-strain curves for flaws located in the weld metal and HAZ, (2) the treatment of strain concentration, and (3) the treatment of residual stress.

Copyright 2012 by ASME

conditions were shown to be a factor of 1.8 for low Y/T material (Y/T=0.76) and over 5.0 for high Y/T material (Y/T=0.95). PHYSICAL PROCESS OF TENSILE STRAIN FAILURE The physical process of flaw growth and tensile strain failure is briefly described here. For ductile fracture, the flaw growth consists of two components: blunting and ductile tearing. As applied stress or strain increases, the flaw first blunts from its initial sharp tip and then the blunted flaw initiates stable ductile tearing. Upon further loading, the ductile tear grows in size, forming a growing flaw with a sharp tip. The initial blunted profile remains behind the newly formed sharp flaw. Further ductile tearing eventually triggers the onset of unstable flaw growth. The evolution of the flaw profile is schematically shown in Figure 2, where CMOD is the crack month opening displacement. The limit state corresponding to the onset of stable flaw extension is termed initiation-control limit state. The limit state corresponding to the onset of unstable flaw growth is termed ductile-instability limit state.

The flaw growth vs. strain measured in pressurized and non-pressurized full-scale pipe tests by Minnaar et al, [22] is shown in Figure 5. Similar to the above observations, it is seen that at a flaw growth of 0.5-0.6 mm, the strains are very close to the final failure strains.

Figure 3 Flaw growth vs. strain history from full-scale pipe tests [29]

CMOD

a ~ 0.65 mm

Figure 4 Crack opening profile vs. CTOD-strain history from full-scale pipe tests [30] Figure 2 Schematic illustration of flaw growth profile as a function of remote applied strain The sequence of flaw growth with respect to the remote applied strain may be examined to determine the influence of flaw growth on the final failure strain (tensile strain capacity). The strain vs. flaw growth history of full-scale tests by Igi et al, [29] is shown in Figure 3. It is evident that the strain at 0.5 mm flaw growth is very close to the strain of leakage. The flaw growth at the point of leakage is much greater than 0.5 mm. However, most of growth occurs near the final leakage point accompanied by a small increase of remote strain. The CTOD vs. strain history measured from multiple fullscale pipe tests from stby and Hellesvik [30] is shown in Figure 4. The amount of flaw growth is about 0.65 mm at the point when the CTOD and strain relation turns nearly vertical, indicating that a small increment of strain causes a large increase of CTOD. The crack profile taken by silicone molding indicates that most of the flaw opening and growth was from blunting. The stable tearing has just begun to initiate.

Figure 5 Crack growth vs. strain history from full-scale pipe tests [22] The overall observations of those tests are that (1) a large amount of flaw growth can be observed at the termination of a test and (2) much of the flaw growth occurs after the flaw growth vs. strain relation turns nearly vertical, indicating that

Copyright 2012 by ASME

the most of the flaw growth has a small influence on the final failure strain or strain capacity. These observations are consistent with general expectations. When the flaw growth rate starts to accelerate, the reduction of the remaining ligament accelerates accordingly. The reduction of the ligament in turn accelerates the flaw growth. This process leads to the flaw instability with only a slight increase of applied strain. FRAMEWORK OF PRCI-CRES TENSILE STRAIN MODELS The tensile strain models developed by CRES for US DOT and PRCI recognize two limit states, initiation-control and ductile-instability control [31,32]. Both limit states use the same crack driving force relations which are fully described by Liu et al., [1]. The significance and determination of the initiation toughness (related to the initiation-control limit state) and resistance curves (related to the ductile-instability control limit state) are described below. Initiation/Apparent Toughness Similitude of Crack-Tip Fields The fundamental basis of fracture mechanics is the existence of the unique relationships between the crack-tip fields and the fracture parameters, such as the stress intensity factor K, J-Integral, and Crack Tip Opening Displacement (CTOD). When such unique relationships exist, the crack-tip fields of a test specimen can be transferred to the structures being assessed through the fracture parameters, since those parameters uniquely represent the crack-tip fields. Validity Limit of Fracture Toughness A unique relationship between the crack-tip fields and the fracture parameters exists when the crack-tip plasticity is wellcontained within the boundary of the specimens and/or structures. For fracture mechanics test specimens, such a unique relationship is enforced by specifying specimen validity limit as expressed in the following format:

Toughness at the Point of Tensile Failure When the crack-tip profiles of large-scale test specimens are observed, the CTOD values associated with the final failures are well above 0.5 mm, often in the range of 1.0-2.0 mm. These values are much greater than the single-parameter valid CTOD value (0.1 mm or less). This demonstrates that the single fracture parameter characterization of girth weld toughness typical of stress-based design is no long rigorously valid for strain-based design. The concept of apparent toughness is therefore necessary to highlight that the toughness representation is not strictly a material parameter anymore. The apparent toughness represents the combined effects of the intrinsic material toughness and the structural behavior. Therefore it should be expected that the value of the apparent toughness may be dependent on the flaw size and other structural features, such as dimensions and even materials tensile properties. Apparent Toughness from Resistance Curves The crack-tip profiles of a flaw at various stages of a toughness resistance curve are shown in Figure 6 [37]. Upon initial loading, blunting occurs at the crack tip. A small sharp crack initiates upon further straining. Continued loading results in growth of the sharp crack.
CTODR
Conventional CTODR
Maximum Load

a = 0

Flaw Growth a

J / 0

cr

L ,

Figure 6 Schematic illustration of crack-tip profile vs. a resistance curve (CTODR)

where L is the specimen ligament size, and cr is 25 for a bend specimen [33] and 200 for a tension specimen with through wall crack [34,35]. The fracture mechanics parameter J may be converted to CTOD through the following relation [36]:

(0.5 0.7)

J . 0

If the ligament size L of a standard CTOD specimen is 25 mm (equivalent to tests done on a 25-mm wall thickness pipe), the maximum valid CTOD is 0.5-0.7 mm for bend specimens and less than 0.1 mm for tension specimens.

Figure 7 Crack-tip profile from a full-scale test specimen

Copyright 2012 by ASME

CTOD From Flaw Growth

Residual CTOD R (mm)

4.0 3.5

(c) BM: X65 High Y/T WM: Evenmatch Flaw: Pipe Body and HAZ

CTOA

3.0 2.5

CTODR

2.0 1.5
1.0
Blunting

as Original Flaw Tip

a

0.5 0.0 0.0 0.5 1.0 1.5 2.0 2.5

Total

3.0

3.5

Flaw growth a (mm)

Figure 8 Crack opening profile related to two contributing parts

Residual CTOD R (mm)

4.0 3.5
3.0 2.5 2.0 1.5 1.0 0.5 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Blunting Total

(d) BM: X65 High Y/T WM: Evenmatch Flaw: WM

Flaw growth a (mm)

Figure 9 Crack-tip profiles from multiple cuts along the flaw length
Residual CTOD R (mm)
3.5 3.0 (a) BM: X65 High Y/T WM: Overmatch Flaw: HAZ

3.5 3.0

(e) BM: X65 Low Y/T WM: Overmatch Flaw: Pipe Body and HAZ

2.5
2.0 1.5 1.0
Blunting

Residual CTOD R (mm)

2.5
2.0 1.5 1.0

0.5
Blunting Total

Total

0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

0.5
0.0 0.0 0.5 1.0 1.5 2.0 2.5

Flaw growth a (mm)

3.5

3.0

Flaw growth a (mm)

3.5
3.0 (b) BM: X65 High Y/T WM: Overmatch Flaw: WM

3.5 3.0

(f) BM: X65 Low Y/T WM: Overmatch Flaw: WM

Residual CTOD R (mm)

2.5
2.0 1.5 1.0
Blunting

Residual CTOD R (mm)

2.5 2.0 1.5 1.0

Blunting

0.5
0.0 0.0 0.5 1.0 1.5 2.0 2.5

Total

0.5 0.0 0.0 0.5 1.0 1.5 2.0 2.5

Total

3.0

3.5

Flaw growth a (mm)

3.5

3.0

Flaw growth a (mm)

Figure 10 Total and blunting part of residual CTOD as a function of the amount of flaw growth

Copyright 2012 by ASME

An actual crack-tip profile from one of the full-scale test specimens is shown in Figure 7 [32]. This profile shows that the increase of CTODR in the resistance curve comes from two parts: (1) blunting of the initial crack tip and (2) growth of the sharp crack. The second part of the CTODR can be estimated through the use of CTOA, or crack tip opening angle, as shown in Figure 8. A sample set of crack-tip profiles with various amount of flaw growth is shown in Figure 9 [31]. The profiles are from different flaws installed in the same test specimen. The contribution of the advancing sharp crack is computed as the CTOA multiplied by the amount of sharp crack advance, as (see Figure 8). The process described above was applied to the full-scale test specimens described in a companion paper [2]. As the measurements were taken directly from crack-tip profile after the termination of the tests, the values obtained are termed residual CTOD. The total residual CTOD and the blunting part of the residual CTOD are shown in Figure 10. It is evident from Figure 10 that the total residual CTOD follows the general trend of CTOD resistance curve. The blunting part of CTODR remains largely constant, even after a significant amount of flaw growth. The increase of CTODR after a small initial growth is largely attributable to the further opening from advancing crack front. The blunting part of CTODR from weld centerline flaws is smaller than that from HAZ flaws. This trend is consistent with the other measures of toughness, such as Charpy, traditional high-constraint CTOD, and low-constraint SENT resistance curve [31]. Determination of Initiation/Apparent Toughness Determination from Crack-Tip Profile The apparent toughness (CTODA) is the initiation toughness which corresponds to the transition from flaw blunting to ductile tearing with a macroscopic sharp flaw. Similar to the conventional CTOD toughness, the apparent toughness (CTODA) can be measured directly from a crack-tip profile which contains finite ductile tearing as shown in Figure 11. There are three important points to consider in the use of this approach. First, a small amount of sharp flaw growth is necessary to ensure that the full blunting has been exhausted. Secondly, the flaw profile should be obtained from structurerelevant test specimens, such as CWP and full-scale specimens. Small-scale specimens, such as SENT, may be used when the transferability of the specimens to the full-scale structure is accounted for (see discussion about transferability in a later section). Thirdly, when the crack-tip profiles are taken from the crack-tip cross-section after the termination of a test, the crack tip opening is reduced from the elastic unloading. The effects of the unloading should be accounted.

s2

s1

CTODA = s1 s2
Figure 11 Determination of initiation/apparent toughness (CTODA) from crack-tip with finite tearing Direct Measurement from Small-Scale Specimen Although the conventional CTOD toughness can be directly measured from the crack-tip profile in well-calibrated interrupted tests [38], this technique is not easy to use as a routine test procedure. Standard CTOD tests are done by measuring load and CMOD. The CTOD toughness is obtained from a set of correlation equations which relate the load vs. CMOD record with the CTOD. Techniques capable of measuring the transition from blunting to sharp flaw growth can be used to determine CTODA. One possibility is using the potential drop (PD) method to determine the transition between blunting and ductile tearing in low-constraint test specimens, such as SENT, shallow-notched SENB, or CWP specimens. In the PD test method a constant current power supply is connected to the test specimen and the voltage across the notch is monitored during the test. The PD response initially shows a linear increase in PD due to plasticity and crack tip blunting. As ductile tearing initiates and the remaining ligament starts to reduce in size the PD trace exhibits an upward swing. The transition between blunting and ductile tearing corresponds to the start of the upswing in the PD trace. The use of PD method to determine initiation toughness in practical test lab environment can be a subject of further research. The transferability of toughness between low-constraint smallscale specimen, such as SENT and shallow-notched SENB, and large-scale structures, such as CWP and full-scale pipe should be considered. Determination from Regular Small-Scale Test Specimens The initiation/apparent toughness may be determined from a number of small-scale test specimens, including Charpy impact energy, standard high-constraint SENB specimens, low-constraint shallow-notched SENB specimens, and low-constraint SENT specimen. The methodology is covered in another publication [32].

Copyright 2012 by ASME

Resistance Curves and Transferability Observation of Test Data One of the key issues in the development of tensile strain models is the transferability of toughness measurement among specimens of different scales. The relative ranking of resistance curves among SENT, curved wide plate (CWP), and full-scale specimens is not consistent in various public domain publications. For instance, SINTEF shows that the ranking of the resistance curves from high to low is full-scale, CWP, and SENT specimens [39]. Cheng, et al, shows that the resistance curves of full-scale and SENT specimens are the same or very close and the resistance curve CWP is below both full-scale and SENT specimens [40]. The transferability of resistance curves between full-scale pipes and SENT was examined in a great detail by Wang, et al. [32]. The process involved the construction of resistance curves from full-scale test data using the residual flaw profiles and the comparison of the resistance curves from the full-scale specimens with those from SENT specimens. The residual flaw profile reflects the flaw profile after unloading at the termination of the test. In order to construct the flaw profile before unloading, the residual CMOD from the residual flaw profile is compared with the clip gage CMOD measured before the unloading. A nearly linear relationship is established between the residual CMOD and the measured CMOD before unloading, as shown in Figure 12. It is evident that the CMOD before unloading is approximately 10% higher than the residual CMOD.
5.0 4.5
CTODR (mm)

3.5 3.0

(a) BM: X65 High Y/T WM: Overmatch Flaw: HAZ

2.5
2.0 1.5 1.0

0.5
0.0 0.0 0.5 1.0 1.5 2.0

Full Scale

SENT (Power Fit)

2.5

3.0

3.5

Flaw growth a (mm)

3.5
3.0 2.5 (b) BM: X65 High Y/T WM: Overmatch Flaw: WM

CTODR (mm)

2.0 1.5 1.0

Full Scale

0.5
SENT (Power Fit)

0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Flaw growth a (mm)

5.0

4.0

(c) BM: X65 High Y/T WM: Evenmatch Flaw: Pipe Body and HAZ

CTODR (mm)

3.0

BM: X65 High and Low Y/T WM: All Flaw: BM, WM, and HAZ

Residual CMOD (mm)

4.0 3.5 3.0

2.0

1.0

Full Scale SENT (Power Fit)

0.0

2.5 2.0

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

y = 0.89x R = 0.91
3.5 3.0

Flaw growth a (mm)

1.5 1.0
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
CTODR (mm)

(e) BM: X65 Low Y/T WM: Overmatch Flaw: Pipe Body and HAZ

CMOD before Unloading (mm)

2.5
2.0 1.5 1.0
Full Scale

Figure 12 Relationship between the residual CMOD and the measured CMOD before unloading

0.5
0.0 0.0 0.5 1.0 1.5 2.0 2.5

SENT (Power Fit)

3.0

3.5

Flaw growth a (mm)

Figure 13 Comparison of resistance curves between fullscale pipes and SENT

Copyright 2012 by ASME

The residual CTOD from full-scale specimens, similar to that shown in Figure 10, was multiplied by a factor of 1.1 to obtain the CTOD before unloading. Figure 13 shows the comparison of resistance curves between the full-scale pipes and the corresponding SENT for four groups of material and flaw locations. Two trends are evident from the comparison. First, the resistance curves of the pipes are generally higher than those of SENT. Secondly, resistance curves of the fullscale pipes are slightly higher than those of SENT in some cases, but significantly higher in other cases. The comparison of the full-scale and small-scale test data demonstrates that the transferability of resistance curves between the full-scale pipes and SENT cannot be assumed in all cases. This observation is generally consistent with the philosophy of DNV-RP-F108 and different from the results presented by Cheng, et al [39]. One of the possible contributors is test protocols used in generating the resistance curves. The SENT resistance curves cited in those references were generated using different test protocols. Furthermore, the data acquisition and post-test data analysis of the full-scale tests can also affect the presented resistance curves. Full details of the processes are not provided in some of those references. Resistance Curves from Micromechanical Modeling The transferability of the fracture toughness from different tests was also studied using damage mechanics as implemented in the GTN model [32]. Three test forms, including full scale pipe (FSP), CWP and SENT were investigated. The same wall thickness of 11.57 mm was used for all specimens. The outside diameter (OD) of the CWP and FSP specimens was 312 mm. The SENT specimen had a square-shapef cross section. The width of the weld and heat affected zone (HAZ) was 6.96 mm and 2.00 mm, respectively. The weld was parallel-sided (i.e., bevel angle is zero) without cap reinforcement. The initial flaw depth was 3.0 mm for all specimens. The initial flaws were semi-elliptic in FSP and CWP specimens with the length of 50.0 mm. The SENT specimen has a constant flaw depth through the thickness. The damage of the material at crack tip during the tests was accounted by the GTN model, which simulates the mechanism of voids nucleation, growth and final coalescence in metals. The values of the damage parameters in the GTN model were taken from a publication by Liu and Wang [41]. Finite element (FE) analysis was conducted using a commercial software package ABAQUS/Explicit. The FE models had three-dimensional eight-node brick (linear) elements. The elements were removed when the failure criteria of the GTN model were reached to simulate the flaw growth.

The equivalent plastic strain around the crack tip is compared among the different specimens in Figure 14 at three levels of flaw growth. The contours are selected at the cross section normal to the flaw surface at the middle point of the flaw length where the maximum flaw growth is expected. The FSP and CWP specimens have almost the same crack-tip deformation. The crack-tip deformation of the SENT specimen is lower. Figure 14 also shows clearly the removal of damaged elements during the flaw growth. A layer of elements at the newly formed surface (from the initial crack tip location) is removed. The CTOD resistance curves from the FEA are plotted in Figure 15. Due to the removing of element layer at the newly formed surfaces, it is not feasible to determine the CTOD based on the crack-tip deformation profile. A representative CTOD was used in Figure 15 as the separation of the crack surfaces 0.25 mm behind the original flaw tip. The figure shows almost identical resistance curves from FSP and CWP specimens. The resistance curve from the SENT specimen is about 20% lower than the other two specimens. The results are consistent with the plastic strain fields of Figure 14. The overall results show that CWP specimens are capable to provide resistance curves similar to those of FSP. The resistance curve of SENT specimens deviates from those of FSP and CWP.
FSP
a 0.51mm CWP

SENT

FSP

a 0.97 mm CWP

SENT

FSP

a 1.91mm CWP

SENT

Figure 14

Equivalent plastic strain field around the crack in FSP, CWP and SENT specimens at different flaw growth

Copyright 2012 by ASME

0.8 Initial flaw depth a0 = 3mm 0.6

4.

CTOD (mm)

0.4

Collect basic design and material information which affect the tensile strain capacity. Some of the key parameters are (but are not limited to): (1) pipe wall thickness, (2) pipe tensile properties, (3) weld tensile properties, (4) weld toughness, (5) pipe dimensional tolerance, (6) girth weld high-low misalignment, (7) target acceptable flaw size, and (8) inspection method. Determine if there are time-dependent degradation mechanisms of the material properties, such as strain aging or hydrogen embrittlement. If present, these mechanisms should be considered. The effects of the material property degradation may be incorporated into the material qualification phases or monitored as a part of the pipeline integrity management program. Conduct a preliminary assessment of the tensile strain response for all postulated failure events. A target category of tensile strain behavior may be selected. More information on the category of tensile strain response can be found in reference [42]. For a small project, in which the material property variations can be well defined, conditions for Category I behavior, i.e., ensuring the postulated failure locations in the pipe body, may be explored. Select an appropriate level of the tensile strain procedures if Category III behavior is targeted. This may be done by balancing the needs for accuracy and the requirements for material property data, flaw size, and construction quality control. The higher the required accuracy, the greater the need for detailed material qualification. Develop linepipe specifications and welding procedure requirements based on the input requirements of the tensile strain models. Conduct material characterization requirements of No. 8. tests per the

5.
FSP

0.2

CWP SENT

0.0 0.00

0.50

1.00

1.50

2.00

a (mm)

Figure 15

Resistance curves of CTOD versus flaw growth from FEA of FSP, CWP and SENT.

6.

OVERALL PROCESS OF TENSILE STRAIN DESIGN A tensile strain design process involves three essential elements: (1) linepipe specification, (2) welding procedure qualification, and (3) tensile strain models. One critical difference between the strain-based design and the traditional stress-based design is that much more material information is needed for the strain-based design. Furthermore, the information must be defined in greater detail and higher precision in the strain-based design than in the stress-based design. Expert advice may be needed in some cases to bridge the gap between the current industry practice and the additional requirements needed for the strain-based design. All phases of a pipeline life, including installation, commissioning, and operation should be considered to ensure safe operation throughout its entire design life. The overall tensile strain design process may involve the following steps. The steps are intended to highlight some of the key considerations. They are, by no means, to be allinclusive. Not all steps are applicable in all situations. 1. Determine the nature of the strain demand. The strain demand could be a one-time event, such as the strain at a fault crossing in a seismic event or accumulative over a service period of a pipeline, such as frost heave and thaw settlement. Strain demand in the area of mine subsidence could be a one-time event or a time-dependent accumulative event. Determine and classify postulated failure events. In some cases, the most critical event might be in the construction/installation phase, such as an offshore installation by reeling. For most onshore pipelines, the operation phase is more critical as the tensile strain capacity is reduced by the internal pressure. Determine a set of target tensile strain capacity levels by introducing appropriate safety factors to the estimated strain demand.

7.

8.

9.

10. Evaluate tensile strain capacity against the target values, taking into consideration the material property variations from No. 9. 11. Conduct confirmation tests of the tensile strain capacity when needed. 12. Develop and implement material property surveillance protocol if time-dependent degradation mechanisms exist and their effects are not fully covered in the material qualification phase. 13. Develop and implement strain demand monitoring systems. When needed, verify the reliability and accuracy of such systems. 14. Develop and implement continuous evaluation and mitigation plans if the strain demand and material properties evolve over time.

2.

3.

Copyright 2012 by ASME

Some of the above steps may be iterated in the design phase or in the maintenance and integrity management phase when more detailed or reliable information becomes available. KEY CONCLUSIONS AND FUTURE WORK A number of conclusions may be drawn from the information presented. 1. The failure of a planar weld flaw under longitudinal tensile loading involves the initial blunting of the flaw, initiation of stable flaw growth, and onset of unstable growth after a finite amount of flaw growth. When the amount of flaw growth is measured against the magnitude of the applied remote strain, it appears that most of the flaw growth observed at the termination of a test occurs after the strain reaches a value close to the final failure strain. In other words, most of the remote strain is attained before the onset of stable tearing (initiation). Since most of the remote strain is attained prior to initiation, tensile strain limit state on the basis of initiation should produce similar tensile strain capacity as the limit state on the basis of ductile tearing instability. Apparent toughness is the initiation toughness at the onset of stable ductile tearing. As such, its value is closely related to the blunting part of a crack-tip profile. The detailed examination of full-scale and small-scale test data and damage mechanics analysis demonstrate that the transferability of resistance curves between the full-scale pipes and SENT cannot be assumed in all cases. More indepth analysis of past and future data is necessary to establish the transferability. Such analysis can only be done if the details of the instrumentation plan, data acquisition method, and post-test data processing procedure are provided for review and verification. 3

Conference, Paper No. 90660, Calgary, Alberta, Canada, September 24-28, 2012. Hukle, M. W., Horn, A. M., Hoyt, D. S., and LeBleu, Jr., J. B., Girth Weld Qualification for High Strain Pipeline Applications, Proceeding of the 24th International Conference on Offshore Mechanics and Arctic Engineering (OMAE2005), June 12-17, Kalkidiki, Greece, 2005. Denys, R., Lefevre, A., and Baets, P. D., A Rational Approach to Weld and Pipe Material Requirements for a Strain Based Pipeline Design, Proceedings of Applications & Evaluation of High-Grade Linepipes in Hostile Environments, Pacifico Yokohama, Japan, November 7-8, 2002, pp. 121-157. Wang, Y.-Y., Liu, M., Chen, Y., and Horsley, D., Effects of Geometry, Temperature, and Test Procedure on Reported Failure Strains from Simulated Wide Plate Tests, Proceedings of 6th International Pipeline Conference, Paper No. IPC2006-10497, September 25-29, 2006, Calgary, Alberta, Canada. Denys, R.M. and Lefevre A.A., UGent guidelines for curved wide plate testing, Pipeline Technology Conference, Paper number Ostend2009-110, Ostend, Belgium, October 12-14, 2009. Wang, Y.-Y., et al., Weld Design, Testing, and Assessment procedures for High Strength Pipelines Summary Report, PHMSA contract No. DTPH56-07-T000005, report No. 277-S-01, http://primis.phmsa.dot.gov/matrix/PrjHome.rdm?prj=225 &s=AF9B5DD9889146979C92C295C416E17C&c=1. 8 Weeks, T., et al., Weld Design, Testing, and Assessment Procedures for High-strength Pipelines - Curved Wide Plate Tests, PHMSA contract No. DTPH56-07-T000005, report No. 277-T-09, http://primis.phmsa.dot.gov/matrix/PrjHome.rdm?prj=225 &s=AF9B5DD9889146979C92C295C416E17C&c=1. 9 Wang Y.-Y., et al., Weld Design, Testing, and Assessment Procedures for High-strength Pipelines - Curved Wide Plate Test Results and Transferability of Test Specimens, PHMSA contract No. DTPH56-07-T-000005, report No. 277-T-11, http://primis.phmsa.dot.gov/matrix/PrjHome.rdm?prj=225 &s=AF9B5DD9889146979C92C295C416E17C&c=1. 10 DNV OS-F101, Submarine Pipeline Systems, October 2010. 11 DNV-RP-F108, Fracture control for pipeline installation methods introducing cyclic plastic strain, January 2006.

2.

3.

4.

5.

ACKNOWLEDGMENT The financial and technical support of US DOT PHMSA, the Pipeline Research Council International (PRCI) and its member companies are gratefully acknowledged. REFERENCES

Liu, M., Wang, Y.-Y., Song, Y., Horsley, D., and Nanney, S., Multi-Tier Tensile Strain Models for Strain-Based Design Part II - Development and Formulation of Tensile Strain Capacity Models, Proceedings of the 9th International Pipeline Conference, Paper No. IPC201290659, Calgary, Alberta, Canada, September 24-28, 2012. Liu, M., Wang, Y.-Y., Horsley, D., and Nanney, S., Multi-Tier Tensile Strain Models for Strain-Based Design Part III - Model Evaluation against Experimental Data, Proceedings of the 9th International Pipeline

10

Copyright 2012 by ASME

12 BS 7910:2005, Guide to Methods for Assessing the Acceptability of Flaws in Metallic Structures, BSI, 2005. 13 Wang, Y.-Y., Rudland, D., Denys, R., and Horsley, D. J., A Preliminary Strain-Based Design Criterion for Pipeline Girth Welds, Proceedings of the International Pipeline Conference 2002, Calgary, Alberta, Canada, September 29-October 3, 2002. 14 Wang, Y.-Y., Cheng, W., McLamb, M., Horsley, D., Zhou, J., and Glover, A., Tensile Strain Limits of Girth Welds with Surface-Breaking Defects Part I an Analytical Framework, in Pipeline Technology, Proceedings of the 4th International Conference on Pipeline Technology, Edited by Rudi Denys, Ostend, Belgium, May 9-13, 2004. 15 Wang, Y.-Y., Horsley, D., Cheng, W., Glover, A., McLamb, M., and Zhou, J., Tensile Strain Limits of Girth Welds with Surface-Breaking Defects Part II Experimental Correlation and Validation, in Pipeline Technology, Proceedings of the 4th International Conference on Pipeline Technology, Edited by Rudi Denys, Ostend, Belgium, May 9-13, 2004. 16 Horsley, D. and Wang, Y.-Y., Weld Mismatch Effects on the Strain Limits of X100 Girth Welds, in Pipeline Technology, Proceedings of the 4th International Conference on Pipeline Technology, Edited by Rudi Denys, Oostende, Belgium, May 9-12, 2004. 17 Wang, Y.-Y., Cheng, W., and Horsley, D., Tensile Strain Limits of Buried Defects in Pipeline Girth Welds, Proceedings of the International Pipeline Conference 2004, Calgary, Alberta, Canada, October 4-8, 2004. 18 CSA Z662, Oil and Gas Pipeline Systems, Canadian Standards Association, 2007. 19 stby, E, New strain-based fracture mechanics equations including the effects of biaxial loading, mismatch and misalignment, Proceeding of the 24th International Conference on Offshore Mechanics and Arctic Engineering, Paper No. OMAE2005-67518, June 12-17, 2005, Halkidiki, Greece. 20 Sandivk, A., stby, E, Naess A., Sigurdsson G., and Thaulow, C., Fracture control offshore pipelines: probabilistic fracture assessment of surface cracked ductile pipelines using analytical equations, Proceeding of the 24th International Conference on Offshore Mechanics and Arctic Engineering, Paper No. OMAE2005-67517, June 12-17, 2005, Halkidiki, Greece. 21 Nyhus, B., stby, E, Knagenhjelm, H.O., Black, S., and Rstadsand, P.A., Fracture control offshore pipelines: experimental studies on the effect of crack depth and asymmetric geometries on the ductile tearing resistance, Proceeding of the 24th International Conference on

Offshore Mechanics and Arctic Engineering, Paper No. OMAE2005-67532, June 12-17, 2005, Halkidiki, Greece. 22 Minnaar, K. Gioielli, P.C., Macia, M.L., Bardi, F., Biery, N.E., and Kan, W.C., Predictive FEA modeling of pressurized full-scale tests, Proceedings of the 17th International Offshore and Polar Engineering Conference, July, 1-6, 2007, Lisbon, Portugal, P3114. 23 Fairchild, D.P., Cheng, W., Ford, S.J., Minnaar, K., Biery, N.E., Kumar, A., and Nissley, N.E., Recent advances in curved wide plate testing and implications for strainbased design, Proceedings of the 17th International Offshore and Polar Engineering Conference, July, 1-6, 2007, Lisbon, Portugal, P3013. 24 Kibey, S.A., Minnaar, K., Cheng, W., and Wang, X., Development of a physics-based approach for the predictions of strain capacity of welded pipelines, Proceedings of the 19th International Offshore and Polar Engineering Conference, June, 21-26, 2009, Osaka, Japan, P132. 25 Kibey, S., Issa, J.A., Wang, X., and Minnaar, K., A simplified, parametric equation for prediction of tensile strain capacity of welded pipelines, Pipeline Technology Conference, October 12-14, 2009, Paper No. Ostend2009039, Ostend, Belgium. 26 Otsuka, A., et al., Effect of Stress Triaxiality on Ductile Fracture Initiation of Low Strength Steel, J. of the Society of Material Science, Japan, Vol. 29, No. 322, pp. 717-723, 1980. 27 Toyoda, M., et al., Criterion for Ductile Cracking for the Evaluation of Steel Structure under Large Scale Cyclic Loading, Proc. Of 20th International Conference on Offshore Mechanics and Arctic Engineering, OMAE 2001-MAT-3103, 2001. 28 Igi, S. and Suzuki, N., Tensile Strain Limits of X80 High-Strain Pipelines, Proceedings of the 17th International Offshore and Polar Engineering Conference, Lisbon, Portugal, July 1-6, 2007. 29 Igi, S., Sakimoto, T., Suzuki, N., Muraoka, R., and Arakawa, T., Tensile Strain Capacity of X80 Pipeline under Tensile Loading with Internal Pressure, Proceedings of the 8th International Pipeline Conference, Paper No. IPC2010-31281, September 27 October 1, 2010, Calgary, Alberta, Canada. 30 Ostby, E., and Hellesvik, A., Fracture Control of Offshore Pipeline JIP: Results from Large Scale Testing of the Effects of Biaxial Loading on the Strain Capacity of Pipes with Defects, Proceedings of the 17th International Offshore and Polar Engineering Conference (ISOPE 2007), Lisbon, Portugal, July 1-6, 2007.

11

Copyright 2012 by ASME

31 Wang, Y.-Y., Liu, M., Long, X., Stephens, M., Petersen, R., and Gordon, R., 2011, Validation and Documentation of Tensile Strain Limit Design Models for Pipelines, US DOT Contract No. DTPH56-06-T000014, final report to US DOT and PRCI, August 2, 2011, http://primis.phmsa.dot.gov/matrix/PrjHome.rdm?prj=200. 32 Wang, Y.-Y., Liu, M., and Song, Y., Second generation models for strain-based design, US DOT Contract No. DTPH56-06-T000014, final report to US DOT and PRCI, August 30, 2011, http://primis.phmsa.dot.gov/matrix/PrjHome.rdm?prj=201 33 McMeeking, R. M. and Parks, D. M., On Criteria for JDominance of Crack Tip Fields in Large-Scale Yielding, Elastic-Plastic Fracture, ASTM STP 668, ASTM, pp. 175-194, 1979. 34 Begley, J. A. and Landes, J. D., The J-Integral as a Fracture Criterion, Fracture Toughness, Part II, ASTM STP 514, ASTM, pp.1-20, 1972. 35 Shih, C. F. and German, M. D., Requirements for a One Parameter Characterization of Crack Tip Fields by the HRR Singularity, International Journal of Fracture, Vol. 17, No. 1, pp.27-42, 1981. 36 Shih, C. F., Relationships between the J-Integral and the Crack Opening Displacement for Stationary and Extending Cracks, Journal of the Mechanics and Physics of Solids, Vol. 29, pp. 305-326, 1981. 37 Liu, M., Wang, Y.-Y., and Long, X., Enhanced Apparent Toughness Approach to Tensile Strain Design, Proceedings of the 8th International Pipeline Conference, Paper No. IPC2010-31386, September 27 October 1, 2010, Calgary, Alberta, Canada. 38 Wang, Y.-Y., Kirk, M. T., and Reemsnyder, H., Inference Equations for Fracture Toughness Testing: Numerical Analysis and Experimental Verification, 28th National Symposium on Fatigue and Fracture Mechanics, ASTM STP 1321, J. Underwood, etc., Eds., Saratoga Springs, NY, June 25-27, 1996. 39 DNV/SINTEF/TWI, Fracture Control for Installation Methods Introducing Cyclic Plastic Strains, Development of Guidelines for Reeling of Pipelines, a JIP performed jointed by DNV, SINTEF, and TWI. 40 Cheng, W., Tang, H., Gioielli, P., Minnaar, K., Macia, M., Test Methods for Characterization of Strain Capacity: Comparison of R-Curves from SENT/CWP/FS Tests, Pipeline Technology Conference, 12-14 October 2009, Ostend, Belgium. 41 Liu, M. and Wang, Y.-Y., Applying Gurson Type of Damage Models to Low Constraint Tets of High Strength Steels and Welds, Proceedings of the 6th International

Pipeline Conference, Paper No. IPC2006-10416, September 25 29, 2006, Calgary, Alberta, Canada. 42 Wang, Y.-Y., Liu, M., McColskey, D., Weeks, T., and Horsley, D., Broad Perspectives of Girth Weld Tensile Strain Response, Proceedings of the 8th International Pipeline Conference, Paper No. IPC2010-31369, September 27 October 1, 2010, Calgary, Alberta, Canada.

12

Copyright 2012 by ASME